critical deposit velocities for lo'w-concentration...
Post on 14-Mar-2020
0 Views
Preview:
TRANSCRIPT
FRnz ENGINEERINGlABORATORY ~..lOR,~R'"
Millard P. Robinson, Jr., A. M. ASeE,.and Walter H. Graf, M. ASCE
ASCE National Water ResourcesEngineering Meeting
January 24-28, 1972 • Atlanta, Ge~fgia"- $0.50
CRITICAL DEPOSITVELOCITIES FOR
LO'W-CONCENTRATIONSAND-WATER MIXTURES
This preprint has been provided for the purpose of convenient distribution of information at the Meeting. To defray, in part, the cost ofprinting, a meeting price of 50¢ to all registrants has been established.The post-meeting price, when ordered from ASCE headquarters will be50¢ while the supply lasts. For bulk orders (of not less than 200 copiesof one preprint) please write for prices.
No acceptance or endorsement by the American Society of CivilEngineers is implied; the Society is not responsible for any statementmade or opinion expressed in its publications.
Reprints may be made on condition that the full title, name ofauthor, and date of preprinting by the Society are given.
I
fi
CRITICAL DEPOSIT VELOCITIES FORLOW-CONCENTRATION SAND-WATER MIXTURES
by Millard P. Robinson, Jr.1, A.M. ASCE, and Walter H. GraPJ, M. ASCE
INTRODUCTION
This study deals with an important aspect of solid-liquid
transport technology in pipelines: The critical deposit velocity, "VC" ,
which separates the "non-deposit" (deposit free) regime from the "de-
posit" regime. This velocity is sometimes also referred to as either
the minimum transport velocity, the deposition velocity, or just the
critical velocity.
The critical deposit velocity of low concentration mixtures
(C ~ 5%) is presently not well-defined, although it is needed for de-
sign application. Pressurized sewage collection lines, most often
transporting low concentration loads, have been shown to be economica1-
1y competitive with conventional means of sewage disposal but in need
of additional design information. There exists an exhaustive list of
Newtonian slurry transport applications which can be found in the
literature. Condolios et al. (5,6,&7), Shen et al. (27&28), Robinson
et a1. (24), Robinson (25), and Graf (17) report the most current state-
of-the-art and economic significance of the critical deposit velocity
determinat ion.
lEnvironmenta1 Engineering Division, Gilbert Associates, Inc., Reading,Penna. (Former Lehigh University Graduate Research Assistant).
2Associate Professor and Director of Hydraulics and Sanitary Division,Civil Engineering Department, Lehigh University, Bethlehem, Penna.
There exist generally two prerequisites in properly designing
a solid-liquid transport system: (1) Consideration of criteria that
will ensure operation in a region of stability, and thus, provide for
safe, uninterrupted transport of solids, and (2) minimization of the
power required to transport the solids, and optimization of system de
sign parameters. The critical deposit velocity relates both of these
requirements in designing a transport system which is both economic
and safe to operate.
The present study continues the investigation of the critical
deposit velocity problem through the use of a modified Froude number
analysis. From a regression analysis of the Lehigh data, correlation
of the tested parameters with different modified Froude numbers is
evaluated, and equations quantifying the modified Froude number re
lationship are determined. The Lehigh data are subsequently compared
with data reported in the literature, and the economic implications of
applying the resulting Lehigh equations in systems design are dis
cussed.
General Remarks on Solid-Liquid Mixture Flow. It is not
within the scope of this paper to exhaustively present the general
theory for flow of solid-liquid mixtures in pipelines. Shen et al.
(27) and Graf (17) have presented comprehensive surveys on the current
state-of-the-art of sediment transport in pipes, and the interested
reader is referred to these texts. It should be noted, however,
that transported solid-liquid mixtures may vary from suspensions in
water of coal, sand, gravel, wood chips, chopped sugar cane, and ashes
to slurries of sewage sludge, polymeric solutions, and concentrated
suspensions.
-2-
Solids suspensions are transported either as "Non-Settling"
(homogeneous) mixtures or as "Settling" (heterogeneous) mixtures. The
distinction between these two classifications has been presented by
Durand (9) and Govier et al. (IS). The present study is concerned
with a "Settling" mixture, which exhibits Newtonian flow characteristics
and is analyzed as a two-phase flow phenomenon. The transport of "Set
tling" mixtures in pipes is qualitatively characterized by several
different regimes of flow. Reference for an explanation of these dif
ferent regimes is again made to Shen et al. (27) and Graf (17).
The variety of flow regimes and their associated solids con
centration graaients is diagramatically presented in Fig. 1, which is
a typical curve of mixture headloss versus mixture velocity. An im
portant distinction is made between the "Deposit" transport regime and
the "Non-Depos it" transport regime. Within the non-deposit regime,
several modes of transport prevail: (1) Pseudohomogeneous flow,
(2a) heterogeneous flow, and (2b) heterogeneous flow with saltation.
Flow in the deposit regime, (4), is described by bed and. dune form
irregularities. Separating the deposit and the non-deposit flow
regimes, (3), is the transition region identified by the critical
deposit velocity, "VC
" One is reminded that these points of division
between different flow regimes are somewhat arbitrary.
TIlE CRITICAL DEPOSIT VELOCITY, "VC
"
Definition and Significance. The transition between deposit
and non-deposit flow regimes is identified by a "critical condition".
In the present investigation, "critical condition" is taken as the
velocity at which particles begin to settle from the flowing medium
-3-
and form a stationary (non-moving) deposit along the invert of the
pipe; this will be called the critical deposit velocity, "VC,::.
At the "critical condition" a deposit-scour feedback mech-
anism transports solid particles in the form of a pulsating bed. Close
to the pipe wall the solid particles are stationary. When this con-
dition is observed, the critical deposit velocity is recorded. Above
this layer of stationary particles the remainder of the bed is sliding.
Other particles shove, roll, and saltate over the moving bed surface,
and some will become completely suspended farther from the wall. The
deposit of solids on the bottom of a pipe is a random phenomenon vary-
ing with local fluctuations of solid and liquid parameters. Within
the same pump-pipe facility, duplication of results is not easily
attainab Ie.
The critical deposit velocity is sometimes referred to as
the limit deposit velocity, by Durand (9) and Sinclair (29), the
sediment limiting velocity, by Gibert (14), the minimum transport
velocity, by Rose et al. (26), or the deposition velocity, by Wasp
et al. (32). It is imperative that a clearly defined "critical con-I
dition" becomes a primary concern in every solid-liquid transport
investigation.
When using data from other "critical condition" studies,
one must be cautious of the following: (a) Some investigators, such
as Blatch (2), Wilson (34), Bruce et al. (4), Thomas (31), Charles
(8), and Shen et al. (28), define a minimum or economic velocity
which corresponds to the minimum headloss required for transporting
a certain concentration of solids. Use of this criterion is in ac-
cordance with how one wishes to define "critical condition". It
-4-
was found in the present and in other investigations that the critical
deposit velocity is not in direct relationship with the minimum head-
loss criterion. Implementation of the assumption that these two cri-
teria are identical is good only for preliminary evaluation. (b) The
critical deposit velocity, approached from the non-deposit regime,
is most often different from the critical scour velocity. To scour
a deposited bed requires usually a greater shear force, thus a higher
flow velocity, than when the same bed is deposited. (c) Some studies
define a transition velocity between saltating and sliding bedload
transport; which is at times mistaken for the critical deposit ve-
locity.
The critical deposit velocity is an important design cri-
terion both for safe operation and for system economics, but it is
often vaguely defined in reports of solid-liquid transport research.
Due to a lack of good definition and reproduceability of results,
it is suggested that a conservative critical deposit velocity be
used (see also Bonnington (3)].
Previous Investigations. Interest in the "critical con-
dition" of solid-liquid transport in pipes was initiated by Blatch
(2) and continued by O'Brien et al. (21), Howard (18), and others.
However, Wilson (34) developed the first relation~hip which
quantitatively dealt with parameters related to the "critical COn-
dition", which is given in its final form as:
PKCV gDV = ss
C f
-5-
(1)
It should be noted that the flow velocity, VC
' at "critical condition"
is defined here for minimum energy gradients. Nevertheless, the
relationship given with Eq. (1) relates parameters which are of im
portance in the critical deposit velocity problem. These parameters
are: C, the solids concentration; vss ' the particle settling veloc
ity; D, the pipe diameter; f, the friction factor indicating flow
resistance; and a correlation parameter, K.
Durand (9) used as the lower limit of his heterogeneous flow
relationship an equation defining the limit deposit velocity, VC
' of
sand mixtures which separates the zones of the regimes with and without
deposit on the pipe bottom, or:
(2)
The parameter, FL
, known as a modified Froude number, varies with solids
concentration, C, and particle diameter, d. This relationship was ex
amined for the transport of uniformly graded material, and later, Durand
et al. (10) report findings for non-uniform material.
Gibert (14) reported on andanalyzed the extensive SOGREAH data
to obtain best-fit curves for Froude number, vc/~, plotted against
solids concentration, C. Subsequent to the study of Gibert (14), Graf
et al. (16) included the effect of relative density, given byJ2 (ss-l)~
- as was similarly done by Durand (9) - and Gibert's best-fit curves
were replotted and are given with Fig. 2. This figure shows the general
trend of results to be remarkably invariant for sand and gravel of par
ticle sizes d ~ 0.37 rom. The curve for this larger material can be
thought of as being a maximum envelope of FL-values. For finer materials,
in the range of d = 0.20 rom and less, there are distinctive variations
in the curves. Condolios et al. (6) report a figure similar to Fig. 2
but only include an envelope curve for graded and mixed sands of
d> 0.44 rom. It is expected (!) that both Gibert (14) and Durand
et al. (10) used the same set of SOGREAH data.
Gibert (14) also discussed a theoretical approach to the
critical deposit velocity problem, considering the "critical con-
ditions" of flow in a conduit irregard less of flow-through geometry,
to be related through the Froude Law of similitude. A discussion of
Gibert's analysis is found in Robinson et al. (24).
Sinclair (29) conducted tests on sand-water, iron-kerosene,
and coal-water mixtures at concentrations up to 20% flowing in 0.5-inch,
0.75-inch, and 1.00-inch pipe. Through a dimensional analysis of the
variables expected to significantly influence the critical deposit
velocity, Sinclair (29) arrives at an equation, such as:
Vmax (3)
where the modified Froude number is expressed with a solids particle
diameter, d85
He observed that the critical deposit velocity reaches
a maximum between 5 and 20% solids concentration, so that the effect
of concentration could be eliminated by using Vmax
instead of VC
.
Sinclair (29) wrote Eq. (3), for d > 1.5 rom (when C does not enter
the problem), as:
Vmax
. I 08'v2gD (ss-l) •
-7-
1.30 (4)
This may be compared with Durand's results, similarly expressed by:
(5)
For smaller particle sizes, Sinclair (29) examines the
relevance of boundary layer theory to the problem, and suggests that
particle diameter, dss
' takes precedent over the pipe diameter, D,
in their relative influence on the modified Froude number. It is
within this smaller range of partie Ie s.izes that the present study
is conducted.
Flow and particle Reynolds numbers have been investigated
for their applicability as criterion in the critical deposit velocity
problem. Spells (30), Charles (8), and studies by Cairns et al., as
reported by Sinclair (29), correlate the Reynolds number with a modi-
fied Froude number relationship. Correlation in these stud~es, however,
is related to the minimum energy gradient criterion.
A modified Froude number relationship apparently presents a
rather good criterion for evaluation of solid-liquid mixture flow
through pipes. Its relationship to other parameters significant in
the critical deposit velocity problem will be re-examined in the pre-
sent study, and exper~ental findings checked against the SOGREAH data.
A Modified Froude Number Analysis. When transporting a solid-
liquid mixture through a closed conduit, one may expect the following
variables to be of importance: (a) Flow Parameters - V, mixture flow
velocity; g, gravitational acceleration; and vss ' particle settling
velocity. (b) Fluid Parameters - p, carrying fluid density; and v,
-8-
kinematic fluid viscosity. (c) Pipe Parameters - D, pipe diameter: e,
pipe roughness: and tan e, pipe slope. (d) Sediment Parameters - Ps'
solids particle density; d, mean particle diameter; ¥s' particle shape factor
(sphericity): (dgO
Id50
), non-uniformity coefficient of grain distri-
bution: and C, (mOVing) volumetric solids concentration.
Proper grouping of variables into dimensionless parameters
was reported in Graf et al. (16) and is re-examined here:
The relative density, (ss-l), comes from (ps-p)/p where Ss = ps/p.
It is expected that the flow Reynolds number, VD/v, does not
I
~I
[V VD d e dgO ]
f .~' (s -1), -, D¥"' D' tan e, d' C'l gD s v s 50
o (6)
playa significant role in this problem, and is thus omitted. Further,
replacing the general flow velocity, V, with the critical deposit veloc-
ity, VC
' and considering the particle shape factor to be unity for natural
quartz grains or already included in the adjustment of non-spherical par-
ticle sizes, Eq. (6) is rearranged and given by:
d e dgO ]D' D' tan e, d' C
50
o (7)
Note that the flow Froude number, V/JiJD, and the relative density, (ss-l),
both given in Eq. (6), were combined in a densimetric or modified Froude
number, vc/~2gD (ss-l): Equation (7) is somewhat similar to relations
proposed by Durand (9), Sinclair (29), and Barr et al. (1).
For a certain relative pipe material roughness, e/D, and solids
grain size distribution, dgO
Id50
' the applicability of Eq. (7) will be
tested in the form of:
-9-
(8)
The left side of Eq. (8) absorbs the tan e argument, and the best trig-
onometric relationship was determined, after fitting data against several
forms, to be:
The left side of Eq. (8) is a modified Froude number. The form of this
parameter, raising both D and (s -1) to the 1/2 power, has been testeds
and shown to be a reliable criterion.
It is felt that without loss~of generality, it may become
frequently important to replace the relative particle to pipe diameter,
diD, by the particle diameter, d, itself. In this instance, the signif-
icance of D is considered to be wholly described in the Froude number.
In the subsequent discussion, data will be presented and com-
pared in the way suggested with Fig. 2.
EXPERIMENTS
Facilities. A three-story, pressurized and self-contained
solid-liquid transport system was constructed, modified from an open-
tank recirculating system. The frequent use of victau1ic couplings
hastened erection and provided flexibility throughout the pipe system.
The experimental facility consists of: (1) A vari-drive
motor-pump assemblage, (2) an adequately flexible pipeline arrangement,
-10-
lI
(3) a sediment feed and removal system, and (4) the necessary measuring
and regulatory devices. Figure 3 schematically illustrates the general
scale of the overall system; a detailed write-up is given in Robinson (25).
Vari-Drive Motor-Pump - The hydraulic horsepower was supplied
from a vari-drive motor-pump assemblage, functioning as the heart of the
system. The pump, furnished by Ellicott, is a single suction centrifugal
type with cast bronze casing and impeller. The suction pipe is 5-1/2
inches I.D., discharge pipe is 4-1/2 inches I.D., and the impeller diame-
ter is 13-5/8 inches O.D. During the operation of the pump, cooling
water is added continuously to the seal on the motor side of the pump,
also providing a lubricating interface. The drive unit is a Westinghouse
(3 phase-60 cyc le-l2S Hp) ''Magna Flow" motor and is regu lated by a vari-
drive control. The driving unit is of the integral type, is water cooled,
and has an adjustable speed range from 100 to 2153 rpm. Along with the
motor there is an operator's station, excitation unit, and a type SL
autostarter. The entire system operates on 208 volts AC.
Pumping efficiency and impeller capacity were not noticeably
altered throughout the l8-month testing period.
Pipelines - From the pump, mixture flow is discharged through
a 6-inch Foxboro Magnetic Flowmeter leading to a horizontal reach of
8-inch pipe. An 8-inch gate valve regulates pump discharge below flow-
rates of 200 gpm. Often times the partially closed valve would cause
difficulty in establishing stable flow conditions when critical flow-
rates occurred in this lower flow range. The solid-liquid mixture is
then lifted to the test-floor elevation in 6-inch pipe. Along the test
length of approximately 40 ft, measurements are obtained, pipe slope is
adjustable, and mixture flow phenomena are visually observed. A 4-inch
-11-
pipe was installed together with its P1exig1ass observation section;
subsequently, a 6-inch pipe and P1exig1ass section were installed.
A strobotac set at a high frequency response aided the observation of
solids flowing through the P1exig1ass section, such that an accurate
description of flow regime was obtainable.
A "Loop System" follows which is employed as a device for
simultaneously measuring mixture f10wrate and solids concentration.
Located atop the balcony-floor elevation between the 3-inch vertical
pipe sections, commonly referred to as the "Riser" and Downcomer",
is the main air-release for the system.
The flow, upon leaving the "Loop System", bypasses a closed
3-inch sediment flush valve and enters a 6-inch vertical pipe, where
sediment is gravitationally fed when an increase in concentration is
desired. Flow continues downward to where a 6-inch gate valve empties
the system and a 2-inch pipeline connects the city water supply. The
system pressure was maintained and water supply assured through use of
a constant pressure control valve (A in Fig. 3) set at 20 psi on the
2-inch supply line. A 2-inch check valve (B in Fig. 3) prevented back
flow to the city supply under excessive system pressures. The circuit
is completed with S-1/2-inch pipe leading to the suction side of the
pump.
Sediment Feed and Removal System - The sediment feeding ap
paratus underwent several adapt ions until an adequate technique was
successfully applied. Sand was supplied to a mixing chamber and
gravitationally fed to the flowing medium. A sediment removal facility
was employed as a time-saving technique for removing the solids or
-12-
undesirable foreign material from the system and preventing discharge
of polluted water to the collection sump.
Measurement and Flow Regulation - The volumetric concentra
tions of solids and the mixture flowrates were determined from "Loop
System" headloss readings. Arrows 1 and 2 on Fig. 3 indicate the
respective locations of "Downcomer" and "Riser" pressure taps, both
with 1.50 m (=59.1 in.) head loss lengths.
Loop readings were repeatedly checked against flow recordings
from a Foxboro Magnetic Flowmeter by means of a Dynalog Receiver mea
suring accuracy to within 1 percent of full scale, throughout the
scale (approximately ± 25 gpm). A Prandtl tube (C in Fig. 3) was
employed to verify both the "Loop System" and flowmeter measurements
of mixture velocities. A Pitot tube sediment-sampling device (D in
Fig. 3) checked the "Loop System" indication of solids concentrations.
Two Venturimeters were investigated for their applicability
as mixture flow measuring devices, the results of which are reported
by Robinson et al. (24). A new 3 x 2 inch Venturimeter (E in Fig. 3)
and an antiquated 4 x 2 inch device (F in Fig. 3) were tested and later
used in checking flow conditions for this particular study.
The mixture head loss length for the test section was 3.60 m
(=141.8 in.), as located at the arrows marked 3. At each pressure tap
location, four holes, 3/32 inch in diameter, were drilled diagonally
opposite about the circumference of the pipe. Brass fittings were
assembled and connected with poly-flo tubing for transmitting the
hydraulic pressure. Manometer fluids were selected according to the
required range of readings. Most often air-water readings were ade
quate, however, a 2.95 fluid-water medium was needed at extreme flow
-13-
conditions. The 50.0 inch manometer scales were graduated in tenths
of an inch, readings to a hundredth of an inch were estimated, and
each reading was converted to feet of water column. Minor manometer
fluctuations always existed, partly due to the uneven distribution
of sediment concentration through the large system and also due to
the effect that concentrated slugs of sediment had on the pump's
capacity for maintaining a constant mixture f1owrate.
F10wrates between 200 and 1000 gpm were regulated by a vari
drive rheostat control, located at the operator's station. The 8-inch
discharge valve controlled lower range f1owrates. Sediment feed rates
were not rigorously monitored, except for an attempt to evenly dis
tribute the sediment throughout the system.
Measuring Techniques. Clear-water calibration of the system
was the initial course of action. The "Loop System" head10ss readings
were then evaluated and checked against flowmeter, Prandt1 tube, and
Pitot tube measurements.
Clear-Water Tests - Tests of clear-water flow were conducted
to determine material roughness characteristics of the 3-inch "Loop
System" pipes and the 4-inch and 6-inch diameter test lengths. Friction
factors, f, were calculated from the Darcy-Weisbach equation, evaluating
manometer head loss readings and Prandt1 tube indication of velocities
over the ranges of Reynolds number indicated in Table 1. Also summarized
are the respective relative roughness values, e/D, and material values,
e, determined from the Moody-Stanton Diagram of friction factors for
commercial pipe. The friction factors for all three pipes fall in the
transition regime. For further determination of friction factors at
-14-
any mixture flow Reynolds number, an explicit solution of the Colebrook-
White equation was used. Evaluation of extensive "Loop System" data re-
quired this type of solution for f.
The Loop System - The "Loop System" developed by Einstein
et al. (11) was used to simultaneously determine the mixture flowrate,
~. and the solid phase concentration, C. The device consists of two
identical vertical pipe sections with opposite flow direction. Pressure
head differences are obtained over these vertical pipe sections, namely,
the "Riser" and the "Downcomer" section, and ~ and C determined from
relationships based on the sum and differences between the two readings.
To expedite the determination of ~ and C from loop head loss
readings obtained while testing, a program was developed and executed
on the CDC 6400 Computer to print out data for plotting readoff charts.
Description of Experiments. A 4-inch and a 6-inch diameter
pipe, each one having a different pipe roughness, as shown in Table 1,
were evaluated. Each was tested separately at different slopes:
Horizontal; a positive slope, tan e = +0.027; and a negative slope,
tan e = -0.060 (geometrically speaking). Two types of solid particles,
described in Table 2, were tested in various combinations with D and
tan e variables, as are listed in Table 3. The mean sand particle
diameters and non-uniformity coefficients, d60
and dsbd6o
' respectively,
were determined from a standard sieving analysis and remained constant
l throughout the testing period. The settling velocities were found from
r' a graph and equation presented after Budryck by Durand (9, p. 100)."
The specific weights of the solids, ss' were provided by the material
suppliers and are listed in Table 2.
-15-
volumetric concentrations of 0.1% < C < 17% were handled at
no=ates ranging from 0.1 cfs (~50 gpm) < ~ < 1.8 cfs (---800 gpm).
The system temperature was recorded for each test run and sometimes
varied from 60"F < T" < 100"F. The effect of temperature on the loop
readings was accounted for.
For a particular test series, the solids are circulated in
a nearly pseudohomogeneous flow condition which ensures uniform dis
tribution of the particles throughout the system. Once conditions
were stabilized, the flowrate, the moving solids concentration, and
the test section headloss readings were recorded; these are compiled
in Robinson (25). A qualitative description of the mixture flow, as
observed through the Plexiglass section, is thereon commented. Flow
rates are decreased to the heterogeneous flow regime, and eventually
to a heavy bedload transport condition in Which most particles are
either rapidly sliding along the invert or saltating into the clear
flow area of the pipe. Subsequent flowrate changes are more finely
incremented. Lowering the flowrate to a velocity at which the bedload
begins pulsating between deposit and non-deposit flow conditions, the
sliding bed thickness builds and there exists no measurable transport
of the bedload particles. In this study, this is the definition of the
critical deposit velocity, VC' The solids concentration corresponding
to that particular Vc is recorded just prior to the critical condition,
when all particles are in transit.
EVALUATION OF EXPERIMENTAL DATA
Nine series of tests were conducted to determine the critical
deposit velocities for varied concentrations of sand and plastic pellets
-16-
l
I
~
transported with water in a pipeline. Although a series of data was
obtained for the transport of plastic pellets, pumping of the plastics
created considerable problems. The system did not yield consistent
results and these data were considered somewhat unreliable. Con-
sequently, they are not reported here [see Robinson (25)J. Most data
were recorded from sand-water tests in a horizontal pipe over a range
of low solids concentration(C < 7%). It is expected that within this
lower range of solids concentration, both the particle diameter, d,
and solids concentration, C, effect the critical deposit velocity
value.
By testing various combinations of solids concentrations,
C, particle diameter, d, specific weight of solids, ss' pipe diameter,
D, and pipe slope, tan e, different critical deposit velocities were
recorded and compared. All experimental data are first ~abulated
and then plotted as mixture headloss against mixture velocity in
Robinson (25).
Critical Deposit Velocities. The critical deposit velocity
data are summarized in Table 4 with indication of run numbers for each
series of tests, the volumetric solids concentrations, the critical
deposit velocities, and two modified Froude numbers. These two modi-
fied Froude numbers are defined in Table 4 and were computed for each
critical deposit velocity. Froude number (I) is the modified form,
after Durand (9), for critical deposit velocities in horizontal pipe
flow; Froude number (II) is introduced to evaluate critical deposit
velocities in sloping pipes as well.
From a preliminary study, plotting Froude numbers (I) and
(II) against solids concentration, C, it was found that Froude
-17-
number (II) best correlates the data, including both horizontal and
sloping flow values.
Correlation of Data. A regression analysis was made to cor-
relate modified Froude number (II) with the following parameters; con-
centration, C; concentration, C, and particle diameter, d; and con-
centration, C, and relative particle size, diD. The third correlation
provided no additional information and is thus excluded from further
discussion. The regression functions take two forms: (1) A least
squares fit of modified Froude number, Fr, with concentration, C,
written as:
Frk:J
k C1 (9)
where k and k are evaluated from logarithmic values of the data over1 2
three different particle size ranges, and (2) a least squares multiple
regression, using Gaussian iteration to fit modified Froude number, Fr,
to both concentration, C, and particle size, d, such as:
Fr (10)
The exponents, k4
and k6
, and coefficient, k3
, are determined for the
different sand particle ranges of data and also for the total range
of sand-water data. An explanation of the multiple regression analysis
and a statistical interpretation of the resulting equations are given
in Robinson (25).
Two regression equations are found to fit the Lehigh data:
(1) Assuming solids concentration, C, to be the only important in-
dependent variable, the best-fit equation is given as:
-18-
Frv
C [1 - tan eJ~2gD (ss -1)'
0.901 CO.10S (11)
The coefficient of correlation is 0.870. (2) Including the influence
of particle diameter, d, the following equation was developed:
FrVc-;:::====. [1 - tan e]
~2gD (ss-l)0.928 CO.106
(12)
where the particle diameter, d, is in mm. The coefficient of cor-
relation is 0.877. Note that the value of exponent ka = 0.016, given
with Eq. (11), is very close to exponent k4
= 0.105, given with Eq. (12).
Further, coefficient k3
0.928 in Eq. (12) differs only slightly from
coefficient k = 0.901 in Eq. (11). This similarity between the coef1
ficients and exponents in Eqs. (11) and (12) is due to the almost
negligible effect of particle diameter, d. Equations (11) and (12)
are compared graphically in Fig. 4, where Eq. (12) is fit with the two
different particle size data.
From the above discussion, it seems reasonable that the use
of Eq. (11) be recommended.
Relative Influence of Tested Parameters - Needless to say,
not all ranges of the parameters, D, d, ss' C, tan e, d90 /d6o
' and e/D,
have been completely investigated and never will be. However, the re-
su1ting regression equations, Eqs. (9) and (10), offer insight to the
relative influence of some of the tested parameters on the critical
deposit velocity.
The influence of solids concentration, C, on the critical
deposit velocity is found in this study to be of primary significance,
-19-
particularly within a low-concentration range of C < 7%. For con-
centrations above 5 to 10%, both Sinclair (29) and Wilson (33) find
that critical deposit velocities decrease with concentration. A
similar observation was made in the present study when concentrations
exceeded 5%.
The pa~ticle diameter, d, has no direct effect on the critical
deposit velocity value within the range of particle diameters tested in
the present study, 0.45 < d < 0.88 rom. However, with suspensions of
fine particles in the range d < 20 rom, it is expected that solids
settling is sufficiently delayed to decrease the critical deposit
velocity. This is reported by Worster et al. (35) and Gibert (14).
Sloping of the pipe, tan S, noticeably altered the location
of critical deposit. Upward sloped flow hastened settling, yielding
a higher critical deposit velocity than would be expected with similar
flow conditions in a horizontal pipe. The opposite was the case for
downward sloped flow. This influence is explained by the effect that
slope has on the tractive shear force developed in the sliding bed.
While thp. Lehigh data provide insufficient evidence that
relative density, s -1, expressed as (s _1)~5, is proportional tos s
the critical deposit velocity, other studies have made this verifi-
cation. Furthermore, the influence of both the grain size distribution,
d /d , and the relative material roughness, e/D, was not determined90 50
due to the lUnited range over which these parameters were tested.
Comparison with Other Data. Particularly important in the
present study is the applicability of the modified Froude number re-
lationship, given with Eq. (8), for low-concentration mixtures, C < 7%.
The strength of the Lehigh data is in the range with 0.10 < C < 2.0%.
-20-
The low-concentration data are mainly responsible for the final form
of the modified Froude number relationship, as given with Eqs. (11)
and (12). In what follows we shall try to investigate as to how other
experimental data compare with the present findings.
Many researchers have reported on sand-water mixture studies,
but from all of these, only the studies by Gibert (14), F~hrb~ter (13),
Sinclair (29), and Durand, Smith, and Yotsurura, as reported by wasp
~ et al. (32), rendered useful data for the present investigation. The
ranges of parameters investigated in these studies are listed in
Table 5, and the data are plotted in Fig. 5 for comparison with the
Lehigh sand-water data given with:
Fr 0.901 CO.10S (11)
Figure 5 together with the Lehigh sand-water data, represented
with Eq. (11), suggest the following trends in the range where C < 5%:
(1) The critical deposit velocity, VC' increases with solids concentration,
C; the increase becomes less evident as the concentration rises to 5%.
(2) For particle sizes, d ~ 0.37 mm, the critical deposit velocity remains
practically unchanged with increase in d. The Lehigh data exhibit this
trend showing particularly good agreement with the other data, and will
give conservative design values. (3) For particle sizes smaller than
d = 0.37 mm, the critical deposit velocity, VC
' decreases with decreasing
d. It is expected that this decrease in Vc levels off for very fine
particles, but the data reported give inconclusive verification of this.
Neither particle size distribution nor the pipe material roughness were
considered to be of importance in this comparison.
-21-
CONCLUS IONS
The.critica1 deposit velocity, VC
' tested in the form of a
modified Froude number, is correlated with other parameters, which
are significant in the solid-liquid transport problem, over the fo1-
lowing ranges: 0.01 ~ C ~ 7.00%, 0.45 ~ d ~ 0.88 rom, 4.00 ~ D ~ 6.00 in.,
-0.060 ~ tan e ~ 0.027, 1.07 ~ dgO/d50 ~ 1.21, and 0.00009 ~ e/D ~ 0.00032.
From a dimensional analysis of these parameters, a modified
Froude number relationship is developed, as given with Eq. (8). The
relationship is tested for sand-water transport; these data exhibit the
following: (1) Agreement with the Gibert (14) curves for particle
diameters, d ~ 0.37rom; (2) the increase in critical deposit velocity,
VC
' becomes less evident as solids concentration, C, rises to 5%;
(3) for particle sizes, d ~ 0.37 rom, the critical deposit velocity
remains practically unchanged with increase in d; and (4) the critical
deposit velocity is higher than the velocity associated with the
minimum head10ss at low concentrations; however, the opposite is
true for C > 5'70.
A regression analysis, made to correlate the Lehigh data,
shows that the modified Froude number is highly dependent on con-
centration, C, affected by particle diameters, d, if d ~ 0.37 rom,
and hardly influenced by relative particle size, diD. The regression
equation which best fits the data and is in reasonable agreement with
data from other sand-water studies, is given with:
vC [1 - tan eJ
J2gD (Ss-l)'0.901 CO.lOS (11)
The Lehigh critical deposit velocity equations give con-
servative values and are presently the only relations available for
-22-
predicting critical deposit velocities for low-concentration sand-
water mixtures. It is recommended that either Eq. (11) or Eq. (12)
be used as a critical deposit velocity design criterion, certainly
within the range of parameters tested in the present study, and
cautiously in ranges of parameters extending outside of the tested
bounds.
ACKNOWLEDGEMENTS
Sincerest thanks are extended to Oner Yuce1 for his un-
sullied partnership throughout the study. The research program
was partially sponsored by the Federal Water Quality Office of the
U.S.D.I. (WP-01478; 11020 EKD) and by Lehigh's Office of Research.
APPENDIX I - REFERENCES
1. Barr, D. and Ridell, J.: "Homogeneous Suspensions in CircularConduits: A Discussion"; Proc. Amer. Soc. of Civ. Engrs.,Vol. 94, PLl, 1968.
2. Blatch, N. S.: '~orks for the Purification of the Water Supplyof Washington: A Discussion"; Trans. Amer. Soc. of Civ. Engrs.,Vol. 57. 1906.
3. Bonnington, S. T.: "Estimation of Pipe Friction Involved in PumpingSolid Material"; Brit. Hydromechanics Res. Assoc., TN 708,December, 1961.
4. Bruce, 101. A., Hodgson, G. W., and Clark, K. A.: "Hydraulic Transportation of Oil-Sand Tailings in Small-Diameter Pipes"; Trans.Can. Instn. of Mining and Met., Vol. 15, December, 1952.
5. Condolios, E. and Chapus, E. E.: "Transportating Solid Materialsin Pipelines"; Chern. Engrg., June-July, p. 93, 1963.
6. Condolios, E. and Chapus, E. E.: "Designing Solids-Handling Pipelines";Chern. Engrg., June-July, p. 131, 1963.
7. Condolios, E. and Chapus, E. E.: "Operating Solids Pipelines"; Chern.Engrg., June-July, p. 145, 1963.
-23-
8. Charles, M. E.: "Transport of Solids by Pipeline"; Proc. of theFirst Intern. Conf. on the Hydraulic Transport of Solids in Pipes,Coventry, England, Brit. Hydrodynamics Res. Assoc., September, 1970.
9. Durand, R.: "Basic Relationships of the Transportation of Solids inPipes - Experimental Research"; Proc. 5th Congo of Intern. Assoc.Hydr. Res., Minneapolis, 1953.
10. Durand, R. and Condolios, E.: "Technical Data on Hydraulic Transportof Solid Materials in Conduits"; Revue de L'Industrie Minerale,Numero Special IF, June, 1956.
11. Einstein, H. A. and W. H. Graf: "Loop System for Measuring Sand-WaterMixtures"; Proc. Amer. Soc. of Civ. Engrs., January, HYl, 1966.
12. Ellis, H. S., Redberger, P. J., and Bolt, L. H.: "Transporting Solidsby Pipe Line: Basic Principles and Power Requirements"; Ind. andEngrg. Chern., Vol. 55, No.8, August, 1963.
"13. Fuhrboter, A.: "Uber die Forderung von Sand-Wasser-Gernischen inRohrleitungen"; Mitt. d. Franzius-Inst., Techn., Hochschule, Hannover,Heft 19, 1961
14. Gibert, R.: "Transport Hydraulique et Refoulement des Mixtures enConduit";· Annales des Pontes et Chaussees, l30e Annee, No. 12,et No. 17, 1960.
15. Govier, G. W. and Charles, M. E.: "The Hydraulics of the PipelineFlow of Solid-Liquid Mixtures"; The Engrg. Journal, August, 1961.
" "16. Graf, W. H., Robinson, Jr., M. P., and Yucel, 0.: "The Transportof Solid Suspensions in Conduits, Part I: The Critical Velocityfor Solid-Liquid Mixtures"; Fritz Engrg. Lab. Report No. 353.1,Lehigh Univ., Bethlehem, Penna., 1970.
17. Graf, W. H.: "Sediment Transport in Closed Pipes"; Hydraulics ofSediment Transport, McGraw-Hill Publishing Co., Inc., New York,New York, 1971.
18. Howard, G. W.: "Transportation of Sand and Gravel in a Four-InchPipe"; Trans. Amer. Soc. of Civ. Engrs., Vol. 104. 1939.
19. Hunt, W. A. and Hoffman, 1. C.: "Optimization of Pipelines TransportingSolids"; Proc. Amer. Soc. of Civ. Engrs., October, PLl, 1968.
20. Lowenstein, J. G.: "Design so Solids Can't Settle Out"; Chern. Engrg.,January, 1959.
21. O'Brien, M. P. and Folsom, R. G.: "The Transportation of Sand inPipe Lines"; Univ. of Calif. Publications in Engrg., Vol. 3,1937.
22. Roberts, R. N.: "Pipelines for Process Slurries"; Chern. Engrg.,July, 1967.
-24-
" "23. Robinson, Jr., M. P., Yucel, 0., and Graf, W. H.: "The Transport ofSolid Suspensions in Conduits, Part II: Modified Venturimeter;A Measuring Device for Solid-Liquid Mixtures"; Fritz Engrg. Lab.Report No. 353.2, Lehigh Univ., Bethlehem, Penna., 1970.
" II24. Robinson, Jr., M. P. and Yucel, 0.: "Discussion of SedimentTransportation Mechanics: J. Transportation of Sediment in Pipes",by H. W. Shen et al., July 1970; March, HY3, 1971.
25. Robinson, "Jr., M. P.: "Critical Deposit Velocities for Low-ConcentrationSolid-Liquid Mixtures", presented in partial fulfillment for thedegree of Masters of Science; Civil Engineering Dept., Lehigh Univ.,Bethlehem, Penna., 1971.
26. Rose, H. E. and Duckworth, R. A.: "Transport of Solid Particles inLiquids and Gases"; The Engineer, March, 1969.
27. Shen, H. W., Karaki, S., Chamberlain,"Sediment Transportation Mechanics:in Pipes"; Proc. Amer. Soc. of Civ.
A. R., and Albertson, M. L.:J. Transportation of Sediment
Engrs., July, HY7, 1970.
28. Shen, H. W. and Wang, W. S.: "Incipient Motion and Limiting DepositConditions of Solid-Liquid Pipe Flow"; Proc. of the First Intern.Conf. on the Hydraulic Transport of Solids in Pipes, Brit. Hydromechanics Assoc., Coventry, England, September, 1970.
29. Sinclair, C. G.: "The Limit Deposit-Velocity of HeterogeneousSuspensions"; Proc. Symp. on the Interaction Between Fluids andParticles, Inst. of Chem. Engrs., 1962.
30. Spells, K. E.: "Correlations for Use in Transport of AqueousSuspensions of Fine Solids Through Pipes"; Trans. Instn. Chem.Engrs., Vol. 33, 1955.
31. Thomas, D. G.: "Transport Characteristics of Suspensions: Part VI,Minimum Transport Velocity in Large Particle Size Suspensions inRound Horizontal Pipes"; Amer. Inst. of Chem. Engrs. Journal,July, 1962.
32. Wasp, E. J., Aude, T. C., and Kenny, J. P.: "Deposition Velocities,Transition Velocities, and Spatial Distribution of Solids in SlurryPipelines"; Proc. of the First Intern. Conf. on the Hydraulic Transport of Solids in Pipes, Brit. Hydromechanics Assoc., Coventry,England, September, 1970.
33. Wilson, K. C.: ""Derivation of the Regime Equations for PressurizedFlow ..• "; Civ. Engrg. Dept. ,"Queen's Univ., Kingston (Ontario),Report No. 51, 1965.
34. Wilson, W. E.: "Mechanics of Flow with Non-Colloidal Inert Solids";Trans. Amer. Soc. of Civ. Engrs., Vol. 107, 1942.
35. Worster, R. C. and Denny, D. F.: "Hydraulic Transport of SolidMaterials in Pipes"; proc. Inst. of Mech. Engrs., Vol. 169/32, 1955.
-25-
Pipe Specification 010 £ Reynolds Nos.(ft)
Loop System:
3 in. commercial steel 0.00004 0.00001 2.48 x 10" to4.77 x lOS
Test Length:
4 in. galvanized 0.00009 0.00003 1.97 x lOs to3.58 x 10"
6 in. black steel 0.00032 0.00016 1.39 x 10" to3.76 x 105
Table 1: Relative Roughness and Material RoughnessValues for the Three Pipe Sizes
Solids Material d d /d s v60 90 60 S ss
(nnn) (ft/sec)
Quartz Sand:
(f0 0.88 1.21 2.65 0.312
(f00 0.45 1.07 2.65 0.189
Table 2: Solid particles Specification
Pipe Diameter, n in. Mean Particle Diameter, Pipe Slope, Symbol(Material Roughness, dao tan e (Fig. 4)
e f't) (Specific- ravity, ss) (Tab. 4)
4 6 0.88 0.45 0 -0.060 0.027(0.00003) (0.00016) (2.65) (2.65 )
'" '" '" 0
'" '" '" 6.
'" '" '" •'" '" '" A
'" '" '" 0
'" '" '" \1
~' '" '" •'" '" '" l'
Table 3: Tested Combinations of Pipe Diameter,Solid's Particle Diameter, and Slope
-26-
• • - • • •
Fr (II)" Vc . [1 - tan 9JV2&D (8 5-1)
Volumetric Critical Modified
RunSolids Deposit Froude
~ Volumetric Critical ModifiedConc~ntration velocity 1'utnber
R Solids Deposit Froude (percent) (ft/sec) (I) (II)un Concentratil?n Velocity Number ~==_._.-
I (percen~) (ft/sec) (I) (II)Series
~G-OOI •
Series G-OI 0 5 .05 2.75 .462 .4626 .10 4.10 .689 .689
.12 3.90 .656 .656 7 .20 4.80 .807 .807
I :.15 4.65 .782 .782 8 .30 5.45 .916 .916
.20 5.10 .857 .8571 .65 5.10 .857 .857
.50 5.35 .899 .899 9 1.00 5.70 .958 .958
I .50. 5.00 .841 .841 10 1.20 5.85 .983 .983
10 .60· 5.80 .975 .975 2 1.50 5.60 .941 .941
11 1.00 6.40 1.076 1.076 3 3.00 6.25 1.051 1.051
I2 l.00 5.50 .925 .925 4 7,00 6.50 L093 \.093
N 3 I. 75 5.75 .967 .967-.J 4 2.00 5.75 .967 .967I Series G-002 ...
5 5.00. 5.95 1.000 1.0001 .05 3.70 .622 .659
Series G-02 '"2 .10 3.90 .656 .6953 .25- 4.50 . 756 .802
.50 4.80 .807 .855 4 .55 5.10 .857 .909
1.00 S.lei .857 .909 5 2.25 5.50 .925 .980
3.00 5.35 .899 .953 6 2.50 5.70 .958 1.016
7.00 5.00 .841 .891Series BS-OOI •
Series BS~Ol 0 I .75 5.85 .803 .803
.80 6.40 .878 .878 2 1.90 6.95 .954 .954
1.10 6.70 .920 .920 3 2.50 7.45 1.023 1.023
3.00 7.25 .995 .9954 5.40 7.95 1.091 1.091
5.00 7.40 1.016 1.016Series BS-003 "Serie;s B5-03 V .75 6.15 .844 .821
1.00 6.40 .878 .8552.00 7.10 .975 .948
2.30 7.60 1.043 1.0153.70 7.50 1.029 1.002
4.80 7.85 1.077 1.0485.00 7.75 1.064 1.035
Table 4: Critical Deposit Velocity DataTable 4: Continued
Sediment pipe I Sediment SpecificSize Size Cenc. Gravity Remarks
----------I--d[;..,]"-----D---,---C----p.!p---'--------. ---_. ------- -- --'-' -.. ---- .-- - _.
0 0.44 2.65 Extensive rangeDurand (1952)* • 2.04 5.90 in. up to 157. sandi of parllmeters
water tested
2.65 Vc obtained fromSmith (1955) * 0 0.18 3.00 in. up to 267- sandi Vc vs. C plot
'Water
-- > 0.37 40.2 to 2 •.65 Best-fit cuoresGibert (14) -0- -0.20 150.0 mm up to 157- sandi on VcJJi,D' v•• C
water plot
0 0.27 2.64F~hrboter (13) • 0.53, 0.88
0.30 mm up to 257- sandi VC is reportedwater
\l 0.232.65
YotBukura (1961) * T 0.59, 1.154.25 in. up to 257- sandi VC is reported
water
A 0.35 0.50, 0.75, 2.61 VC obtained from.Sinclair (29) .4- 0.68 1.00 in. up to 207- sandI Vc VB. C plot
water
*Reported in WaBp et al. (32)
Table 5: Range of Parameters of the Data Reported by Other Investigatorsfor Sand/Water Mixtures; Data are Plotted 1n Fig. 5.
i (log)m
MixtureHead Loss
1.0 -
0.1
0.01-
~ IDeposit
Regime
10Non-Deposit
Regime
Vm
(log)
MixtureVeloc:i.ty
Fig. 1: R~g1mes of Flow
-28_
6" Flush Valve toCollection Sump
2" City WaterSupply
6" CouplingSleeve
125 H.P. varidrlveMotor- Pump Assemblage
" o{3 Loop ~\O
System o~i ....\o~ff-~'!:'l>
Plexiglas- ,,\e"\,.,..cObservation-Section
17.5'
cW:'di~rn ,..,and Removal
.L':. Devices
Fig. 3. Solid-Liquid Transport Test System
2.0 Vc
/2gD (ss -1)
Adopted from Gibert (14)
_ Sand of d ~ 0.37 mm
1.5 ___ Sand of d = 0.20 mm
1.0~ --------------
0.5
o 10.0Fig. 2: Modified Froude Number versus Concentration;
Particle Diameter as Parameter
-29-
.d=0.88mm _ 0.105 0.056d=0.45 mm}Fr - 0.928 C· d
c[7.]
.-
LEGEND(see also Table 3)
5 ,
~\-r. 0.106
Fr'" 0.901 C
o
o.500~·:·~-----------I
I
10001 -.,.__-
1.0 s.o •• 0 4.0
Fig. 4: Best-Fit Equations for Lehigh's Sand-Water Data OnlYi Modified Froude Numberversus Concentration, Particle Diameter as Parameter
"• f- Vc tan e] Lehigh •[1 -V2gD (s -1)' Sand-Water \
s D. Data
~• -- ", I0
~ ~ - ~-:-- .'V
~.-~-I
I_r0 __ 0 -
t// _0 -0-0
I• I-
__ 0'-
t ....-- 0 0(j)
0 __ d > 0.37 rnm
IGibert
-0- d ;;; 0.20 II1IIl60
I
Durand 0
I
Smith 0II II
LEGEND Fuhrboter 0(see also Table 5) Yotsut,ura \:l
'7 Sinclair 6. C [1.J
I I I
0.7
o.
1.0
1.'
'.0 ••0 4.0 6.0 •• 0
Fig. 5: Modified Froude Number versus Solid 5 Concentration, Particle DialI'.eteras Parameter (Data. from Sand-Water Mixture Studies)
-30-
top related